RESEARCHES IN MAGNETISM. 
Gl 5 
equal than before to what it would be with a load actually on the wire. In the 
present case, to apply and remove 18-| kilos, has raised the magnetic susceptibility 
above the value it had after the wire was demagnetised by reversals. The action is, 
in fact, analogous to the augmentation of magnetism which occurs (through the 
influence of hysteresis) in a magnetised wire which has been tapped, by applying 
and removing a load (see §§ 78, 85). 
Here there is no visible magnetism for the stress to act on, but we find evidence 
of the fact that a load has been applied and removed in the augmented state of 
susceptibility in which it leaves the wire. 
Moreover, just as in the more obvious manifestations of hysteresis already dealt 
with, the effect of vibrating is to remove in great part the traces of previous 
operations. The augmented susceptibility, which is found after loading and unloading, 
disappears if we vibrate the wire before magnetising, just as when a wire hanging in 
a magnetic field is vibrated and then loaded and unloaded, the abnormally raised 
value of the magnetism thereby reached is reduced by again vibrating (§ 85). 
§ 99. I was thus led to conceive of the magnetic susceptibility of iron—a quality 
measurable only by magnetising, but easily thought of as existing apart from any 
actual magnetisation—-as changing with changing stress in a manner which involves 
hysteresis, even when the changes of load take place in the absence of all visible 
magnetisation. Its value depends on previous as well as on actually present loads. 
Now the general relation of susceptibility to stress, as tested by magnetising under 
constant load, resembles that of magnetism to stress when the stress is changed in a 
constant field (cf. figs. 48 and 42), and therefore it seemed probable that the changes 
of susceptibiiity which occur when we load and unload a demagnetised wire (and 
which we now know to be characterised by hysteresis) would, if we could render 
them visible, be found to follow curves very much like those of figs. 31-42. In 
those curves we found a very notable difference between values reached with, say, 
3 kilos., on the on and off branches, and this led me to expect a similar kind of 
difference in the susceptibility of a wire when, after demagnetisation, it was treated 
in these two ways :— 
(1) Load to 18'5, and unload to 3. 
(2) Load to 18’5 ; unload to 0 ; load again to 3. 
The susceptibility in the first case was likely to be greater than in the second, 
except perhaps when the magnetism approached saturation. Further, it seemed 
reasonable to expect that the difference between these cases would be very nearly 
eliminated if in each case the wire was vibrated before the curve of magnetisation 
was taken. 
§ 100. That these expectations were completely fulfilled the following experiment 
(of May 13, 1882) will show. The wire used was the same as before. Curves of 
magnetisation were taken after the several modes of treatment set forth in the 
following table, which also gives the recorded observations of current and magnetism:— 
4 k 2 
